Cargando…

Graph Neural Networks for Maximum Constraint Satisfaction

Many combinatorial optimization problems can be phrased in the language of constraint satisfaction problems. We introduce a graph neural network architecture for solving such optimization problems. The architecture is generic; it works for all binary constraint satisfaction problems. Training is uns...

Descripción completa

Detalles Bibliográficos
Autores principales: Tönshoff, Jan, Ritzert, Martin, Wolf, Hinrikus, Grohe, Martin
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Frontiers Media S.A. 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7959828/
https://www.ncbi.nlm.nih.gov/pubmed/33733220
http://dx.doi.org/10.3389/frai.2020.580607
_version_ 1783665036049252352
author Tönshoff, Jan
Ritzert, Martin
Wolf, Hinrikus
Grohe, Martin
author_facet Tönshoff, Jan
Ritzert, Martin
Wolf, Hinrikus
Grohe, Martin
author_sort Tönshoff, Jan
collection PubMed
description Many combinatorial optimization problems can be phrased in the language of constraint satisfaction problems. We introduce a graph neural network architecture for solving such optimization problems. The architecture is generic; it works for all binary constraint satisfaction problems. Training is unsupervised, and it is sufficient to train on relatively small instances; the resulting networks perform well on much larger instances (at least 10-times larger). We experimentally evaluate our approach for a variety of problems, including Maximum Cut and Maximum Independent Set. Despite being generic, we show that our approach matches or surpasses most greedy and semi-definite programming based algorithms and sometimes even outperforms state-of-the-art heuristics for the specific problems.
format Online
Article
Text
id pubmed-7959828
institution National Center for Biotechnology Information
language English
publishDate 2021
publisher Frontiers Media S.A.
record_format MEDLINE/PubMed
spelling pubmed-79598282021-03-16 Graph Neural Networks for Maximum Constraint Satisfaction Tönshoff, Jan Ritzert, Martin Wolf, Hinrikus Grohe, Martin Front Artif Intell Artificial Intelligence Many combinatorial optimization problems can be phrased in the language of constraint satisfaction problems. We introduce a graph neural network architecture for solving such optimization problems. The architecture is generic; it works for all binary constraint satisfaction problems. Training is unsupervised, and it is sufficient to train on relatively small instances; the resulting networks perform well on much larger instances (at least 10-times larger). We experimentally evaluate our approach for a variety of problems, including Maximum Cut and Maximum Independent Set. Despite being generic, we show that our approach matches or surpasses most greedy and semi-definite programming based algorithms and sometimes even outperforms state-of-the-art heuristics for the specific problems. Frontiers Media S.A. 2021-02-25 /pmc/articles/PMC7959828/ /pubmed/33733220 http://dx.doi.org/10.3389/frai.2020.580607 Text en Copyright © 2021 Tönshoff, Ritzert, Wolf and Grohe. http://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
spellingShingle Artificial Intelligence
Tönshoff, Jan
Ritzert, Martin
Wolf, Hinrikus
Grohe, Martin
Graph Neural Networks for Maximum Constraint Satisfaction
title Graph Neural Networks for Maximum Constraint Satisfaction
title_full Graph Neural Networks for Maximum Constraint Satisfaction
title_fullStr Graph Neural Networks for Maximum Constraint Satisfaction
title_full_unstemmed Graph Neural Networks for Maximum Constraint Satisfaction
title_short Graph Neural Networks for Maximum Constraint Satisfaction
title_sort graph neural networks for maximum constraint satisfaction
topic Artificial Intelligence
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7959828/
https://www.ncbi.nlm.nih.gov/pubmed/33733220
http://dx.doi.org/10.3389/frai.2020.580607
work_keys_str_mv AT tonshoffjan graphneuralnetworksformaximumconstraintsatisfaction
AT ritzertmartin graphneuralnetworksformaximumconstraintsatisfaction
AT wolfhinrikus graphneuralnetworksformaximumconstraintsatisfaction
AT grohemartin graphneuralnetworksformaximumconstraintsatisfaction